Guilty as charged

Hmm, interesting. Somewhat compelling, but:

• it's a rather small (n=38) Chinese pilot study
• the effect on the sleep latency is sizable (a latency decrease from 31±14 to 18±12 minutes, effect size of 0.85), but there's no effect on actual sleep duration.
• the sleep measurements were subjective (sleep diaries, not actigraphy)

I'm also a bit concerned why it's the only study with this methodology in this later meta-analysis - all of the other "behavioral intervention" studies in it experiment with stuff like "extended time-in-bed". In other words, there seems to not have been any followup or replication of this study.